#1 - The Vote
|
|
Maine hasn't elected a new governor with majority support in over 50 years. A 2016 citizen referendum proposes a solution, and a new website brings the power of social choice theory to the people.
March 7, 2017 Like our theme song? Wintergatan made it.
EPISODE LINKS
|
SUBSCRIBE
Subscribe to the show here:
Or anywhere! We're not picky. |
"The vote" full Transcript
JAMES SIMENC: I’m James Simenc, and this is Digits.
[theme music]
JAMES: Hey, welcome everybody to Digits, a show about numbers and us. Now I’ve always fascinated by numbers because they’re everywhere. They’re this tool that we depend on for pretty much everything we do. I mean, they allow us to quantify and make sense our world, and build machines and economies and civilizations. At the end of the day that’s what this show is: it’s me, with a microphone, following that curiosity and investigating stories about numbers and stories that numbers tell about us.
So, when I first started working on the show I didn’t expect to lead with an episode about politics. But then, long story short, um…America 2017. So here we are. We’re in the middle of this crazy, topsy-turvy political moment.
Now, in the wake of last year’s presidential election I was especially interested in conversations about process. About the systems we use to elect our political leaders. And I came across this story from last November that I hadn’t heard anything about at the time. It’s a story about the state of Maine, where a proposal was on the ballot seeking to institute an entirely new voting system in an effort to solve a decades old problem.
KYLE BAILEY: In Maine in nine of the last 11 elections for governor the candidate who won received less than a majority vote, so less than 50% of the vote, and in five of those races, so almost half of our governor elections, the candidate who won received less than 40% of the vote. And so over 60% of voters were voting for someone else.
JAMES: In 1974 James Longley won with 40% of the vote, 1986 John McKernan Jr. 40% of the vote, 1994 Angus King, 35%, 2006 John Baldacci 38%, 2010 Paul LePage 38%.
When it’s possible to win with less than 50%, less than even 40% of the vote, you can have a winning candidate that is actively opposed by 60% or more of the electorate. That is a problem.
Now this result was possible because of Maine’s voting system—called “First Past the Post.” That’s the simple, traditional method we’re all familiar with, where everybody picks their favorite candidate, you count the ballots, and whoever received the most votes wins. Which is fine with only two candidates. But once you add a third and a fourth and a fifth, all of a sudden you can win with a smaller and smaller plurality. In the extreme, it’s possible to win with only two votes if everybody else only gets one.
NISARG SHAH: So my name is Nisarg Shah, and I am a postdoctoral fellow at Harvard University’s center for research on computation and society.
JAMES: It turns out there is an entire field of study about voting and making collective decisions. It’s called social choice theory. That’s where Nisarg comes in. He’s part of the team behind the new website RoboVote.org.
NISARG: So the main idea behind RoboVote is that, look, political elections, everybody knows that they are flawed. But it’s very difficult to get a political reform. It’s very difficult to change the way an election happens at the national level. But the nice thing about RoboVote is that since we have the flexibility of deploying the best algorithms on the website, we can draw on decades of research in the AI community and we can deploy the best algorithms that people can use in their daily life. So at least at that level we can make an impact that we cannot hope to make in the political election scenario in the foreseeable future.
JAMES: RoboVote is designed to help small groups make optimum decisions, things like which company to invest in, which restaurant or movie to go to as a group, which destination to choose for a family vacation.
We make these kinds of decisions all the time. For example just the other day actually, I was out with a group of friends, like 10 of us, and we were trying to decide where to go to dinner. We stood around for half an hour on the sidewalk, totally indecisive, and finally somebody got fed up and said you know what, whatever, I’m going to the Thai place across the street. Now we were all hungry enough at that point not to care much, so we all went to the Thai place across the street. Was that the best choice for the group? Who knows. I’m not sure that “strongest voice wins” is a particularly great system. But the point is people are generally pretty bad at making that kind of decision in an optimum way.
RoboVote on the other hand is great at it. Here’s how it works:
NISARG: So what RoboVote would do is it would ask each participant to rank the different cuisines. So for example you might say that I want to go for pizza, maybe then Indian food, and maybe then Thai food. And I might have a different preference.
JAMES: The idea of ranking is at the core of RoboVote’s process. It’s all about information. If everybody only states their first choice then you know nothing about their preferences among the remaining options. And that matters because of a fundamental truth of collective decisions:
NISARG: There is no one right outcome, there is no one thing that can make everybody happy. Arguably some people are going to be happy with the final choice and some people aren’t.
JAMES: You can’t please everyone. It’s just not possible. So you need a different objective. Here’s where the rankings come in. Instead of thinking about each person’s reaction as binary, as either happy or unhappy, RoboVote sees happiness as a continuous spectrum. Now we can have a new goal. It’s not just about making the most people happy. It’s about making everyone as happy as possible.
If every user ranks their preferences RoboVote can see that, okay, your 1st choice will make you happiest, but your 2nd choice will make you happier than your 3rd, your 3rd than your 4th, and so on. And it can use that information to make a more nuanced decision.
NISARG: So each of us would tell RoboVote’s algorithms our preferences, but it doesn’t know these exact numbers. So what it does is it tries to imagine all the possible ways in which you can assign a number to these different three types of cuisines that are consistent with your ranked preference. And it does that with each participant.
JAMES: RoboVote doesn’t ask the user to specify exact values of happiness for different options because, well, that’s kind of hard to do. But ranking is easy. And the algorithm does the heavy lifting.
You can think about it like this: say we have 10 units worth of happiness to go around. I’m going to call them happions—that’s not a real thing, but if it catches on you heard it here first—so 10 happions. If I ranked Pizza 1st, Indian 2nd, and Thai 3rd, the algorithm says, okay, how can we split up these happions without contradicting the ranking? If pizza’s a strong favorite maybe it gives seven, while Indian only gives two, and Thai one. Or maybe, if I actually like the options pretty equally, it could be something like pizza 3.5, Indian 3.3, and Thai at 3.2. That’s a very different picture, but we haven’t broken any of the rules. There are an infinitely many possibilities. RoboVote looks at all of them, the entire range--
NISARG: It looks at what’s the possible sum of happiness that this option can derive, and then finds the option that would have the greatest sum of happiness no matter what these underlying utilities are. So it gives the greatest sum of happiness to the group even in the worst case.
JAMES: The worst case. That’s where the magic happens. For every outcome, it asks: what is fewest number of Happions the group could end up with? Then, whichever option has the best worst case wins.
NISARG: So they can authoritatively tell this group that, look, after looking at all your preferences, it seems that pizza is the best cuisine for you to go for. And what’s nice is that we have tested these algorithms on real data where people reported their actual happiness value for different options, we converted them to rankings, and we saw how these algorithms performed compared to other algorithms that worked with these rankings. And the actual level of happiness that these algorithms were able to achieve was significantly greater than the other algorithms.
JAMES: So aside from these kind of small scale, practical applications, what do you hope people will take away from RoboVote?
NISARG: So…I guess one outcome that we are trying to hope for is to convey to people that, look, when you have a question that you want to solve, when you have a group decision that you want to make, there are some objectively nice methods that you could use. And these methods are derived in decades and decades of research on artificial intelligence. And let’s not be afraid of them. Let’s rather embrace them, because they can make each and every of all micro decisions better in our daily life.
[“Travel Maine” commercial plays]
COMMERCIAL: This fall, make a brilliant getaway…to Maine!
JAMES: And now we arrive back in the land of lobsters and lighthouses and troubled gubernatorial elections…the state of Maine.
[Skype ringing]
KYLE: This is Kyle Bailey.
JAMES: Hi Kyle how you doing, this is James.
KYLE: Hi James how are you?
JAMES: Doing great, doing great…
JAMES [narration]: Now Kyle was the campaign manager for Maine’s 2016 Question 5 referendum.
KYLE: I moved to Maine in 2010 and experienced two of the nine elections in the last 40 years in which a candidate for governor won without a majority and just watched it all kind of fall apart, have the last three or four weeks of the election be entirely about strategic questions, and have nothing to do with issues or values or vision, and all be about I don’t want to waste my vote, who’s the spoiler, all that stuff. And I just thought I had to take some action and be involved and jumped in with this effort
JAMES: Okay, so…let’s start here. What exactly is Question Five?
KYLE: So Question Five was a citizen initiative on the ballot in 2016 in Maine that asked voters if they wanted to enact ranked choice voting to elect Maine’s United States senators, members of congress, governor, state senators, and state representatives.
JAMES: “Ranked choice voting”, also called “instant-runoff voting.”
JAMES [interview]: Can you walk us through how ranked choice voting works?
KYLE: So if there are more than two candidates in a race, which is very common in our state, you won’t have to just simply choose one, you’ll be able to rank them from your favorite to your least favorite. So your 1st choice, your 2nd choice, and so on. With ranked choice voting you can rank as many or as few candidates as you like. And on election night the town clerk still adds up all the votes to see, in this case, who is everybody’s first choice. Not who is your strategic choice based on polling but who do you actually think would be best for the job.
And if somebody gets a majority—50% plus one—of those first choices they win, right, game over, we’ve already reached a majority winner. But if no candidate has an outright majority of first choices then we look to see which candidate is in last place, who has the smallest pile of ballots in their favor? They’re eliminated. And we have a runoff between the top vote-getters. And let’s say it’s a three-way race, right? No candidate gets majority, the candidate in 3rd place is eliminated, we’re going to have a runoff between the remaining two. But instead of having to come back to the polls four weeks later with four more weeks of negative campaigning, millions of dollars spent on more political TV ads trashing and burning one another, disenfranchising overseas voters including men and women who are serving our country in military and in uniform—you could have an instant runoff.
And you do that with a ranked ballot, so you know who everyone’s first and second choice is. So those voters who liked the candidate who came in last place? We actually look at their ballot to see who is their second choice, who did they like between the remaining candidates? We pick up their ballot and move it to one of the piles for one of the top vote getters, and at that point someone will cross that majority threshold and win. So it’s a way of doing runoffs that’s more efficient, more cost effective, without the downsides of having a whole new election bringing voters back to the polls.
JAMES: Now, Ranked Choice Voting is very simplistic compared to the algorithms used by RoboVote. There are no calculations of happiness, no complex optimization techniques. But the fundamental premise is the same: if you get a more detailed ranked opinion from the voters, you can leverage that extra information to make a better decision, one that more clearly reflects the consensus opinion of the electorate.
And that’s important. Coming to a consensus. I mean, that’s the whole point of a Democracy, right? Enacting the will of the majority. And when that isn’t happening, it causes problems.
KYLE: Nowhere has that been the case truer, longer, more intensely than in Maine. We haven’t elected a governor in our state to their first term with majority support, with a mandate to govern, since 1966. In the state of Maine. So it’s pretty extraordinary, it goes back a long way, and over the decades it has led to increasing dysfunction in our state directly attributable to the fact that we elect governors—and this is just one example—but governors who don’t have a mandate to govern and lead and bring people together, and before they’re even sworn in we get bumper stickers saying, well, 61% of people didn’t vote for you, you’re not legitimate. Or the opposing legislative party says, well, we got a higher percentage of votes to elect our party in the legislature than you did, governor of an opposing party, we’re going to block everything that you want to do. And so the efforts to delegitimize the authority and mandate of that candidate whether or not they would have had a majority with a runoff system is not on the table for us, because we’re absent that. And so it creates real governing obstacles. And ranked choice voting will help to alleviate that. And make it so you cannot say, well, you didn’t get a majority. You very clearly did. You had a runoff process to get there, but this is actually the will of the people, is to go in this direction. So legislature and governor, you guys gotta sit down and learn how to work together.
JAMES: But it’s not just about majority rule. It’s also about improving the quality of political discourse during campaigns.
KYLE: You also have problems where one or more candidates are labeled as spoilers and conversations become all about I don’t want to waste my vote, who’s the spoiler? Instead of who has the best education policy? Who do I agree with on health care? You know, what’s this candidate’s vision for our state? We see that more and more in Maine in races up and down the ballot, but certainly prominent in our governor’s election and it’s such a disservice I think to our voters and to our state.
JAMES [interview]: Yeah, sure and that of course was something that was reflected heavily in our national election.
KB: Yeah, I mean you had a race this year where there were obviously two major party candidates but also a green party candidate, a libertarian party candidate, and others, and there was a lot of conversation about “don’t waste your vote” quote unquote for Gary Johnson or Jill Stein. You know, you might help elect the candidate you like the least. And you actually saw that messaging come from the two big party candidates too, encouraging their supporters not to waste their vote on a third party candidate and that certainly was a part of the conversation and people were talking about having to vote for a lesser of two evils. And so with ranked choice voting you don’t have to do that. You have the freedom to vote for the candidate you like the best without worrying you might help elect the candidate you like the least, knowing that if your favorite candidate can’t win and no candidate gets an outright majority your vote’s not wasted, it counts for your second choice.
JAMES: Now Maine would be the first state to implement this system, but there are a handful of cities in the US that already use ranked choice voting. You have San Francisco, Minneapolis and St. Paul, Cambridge, and Portland in Maine to name a few. Have we seen these kinds of positive effects on the politics of those cities?
KYLE: Well you’ve definitely seen—I’ve heard lots of stories from mayors, let’s say, the mayor of Minneapolis, Betsy Hodges, who in a pretty progressive city, she’s a progressive democrat, she ran a grassroots campaign for mayor of Minneapolis the first time. She knocked on more doors. She asked voters, “If I can’t be your first choice can I be your second choice?” She reached out, listened to more voters, talked to more voters, heard more points of view, and by the way her chief opponent said “don’t rank any candidates, just vote for me,” which was the biggest scandal that entire election because now voters had the power to express their opinions about more than one candidate and this candidate was telling them not to do so. And she ran a positive campaign, she was outspent three to one, but she worked hard, she talked to more voters, she built a coalition, and she won.
And she talks about how she governs as a progressive mayor in a pretty progressive city. Just a recent example there was an agenda that she was advocating for around expanding a set of issues that were important to the labor community, the labor movement, and there was a point where the business community said you know what, we weren’t even brought in on this, we don’t feel engaged. And she said okay, we’ve gotta stop, we’ve gotta go back and bring them to the table so they’re part of the conversation. It’s not that everybody will get what they want, but we do need all the stakeholders to participate in that. And she says that that type of governing is a direct result in part of her personality and how she operates, but also because of her experience on the campaign trail, and having to build majority coalitions and bring people with different viewpoints together. Which is exactly what you want in a governing process.
When you can win as we have seen in Maine with 35 or 37 or 38% of the vote, you don’t really have to talk to a lot of people who disagree with you on substantive issues. You can win by energizing your base, turning them out to vote, and beating the heck out of your opponents to push them just below your 35%. And when you govern you can just dip back into that well in terms of who are you accountable to, because if that’s who you needed to get elected, and who you needed to get reelected, then you’re gonna dance with them that brung you. And that’s a real problem when candidates can win with 37% of the vote or 35% of the vote. We need candidates who can win a majority and bring people together who have different viewpoints to build a winning coalition in a campaign and a governing coalition once they get to the state capitol or Washington DC.
JAMES: So the goal there is improved civility in politics.
KYLE: That’s definitely part of it. I mean, we know that voters and candidates in places that have ranked choice voting report significantly less negative campaigning, so not just voters but candidates saying I got less mud slung at me. So why is that? Let’s—you know, Portland Maine is a great example. The first time we had ranked choice voting we had 15 candidates. The second time we had three. Both the winners and the losers in those elections are supporters of ranked choice voting. Those two chief candidates in both those races had also served in the Maine legislature for a number of years. So they’d won and lost under a first past the post, this-is-the-way-we’ve-elected-people-in-Maine-for-a-long-time-system, and they talked about what that experience was like running for the legislature. So if they’d see a lawn sign for their opponent they’d skip that door and go to the next door. It’s just not a strategic use of their time in a political campaign. But when they ran for mayor with ranked choice voting and they were out knocking on doors, if they saw a lawn sign for one of their opponents they wouldn’t skip that door, they’d knock on it and talk to that voter. Because if they couldn’t be their first choice they wanted to be their second choice. They had to listen to that voter, try to find common ground, talk about the issues, try to figure out where they might jive together, and that candidate in search of second choice and maybe even third choice rankings in a crowded race. And that candidate who had that conversation who maybe wasn’t that voters first choice, maybe their second choice, knows that if they go after the candidate who was that voter’s first choice? If they trash and burn them in a mailer, if they have negative radio ads attacking them? The voter who liked the other candidate the best is way less likely to rank that other candidate who was negative as their second choice. Maybe not even rank them at all. And so negative campaigning can backfire when voters have the power to express their opinions about more than one candidate, and even punish candidates who engage in really negative campaigning.
Under the current system where you can say—and we did see this in the presidential election—vote for me because you hate them and this person over here can’t win? That goes away with RCV. You actually have to as a candidate—you can’t just say vote for me because you hate them and they can’t win—well who says they can’t win? They’re on the debate stage, they can participate without being told to sit down and shut up, they’re a spoiler, raise the issues, try to attract first choice and second choice rankings—it changes the conversation and the tone of campaigns to be much more positive and issue oriented so candidates are held to more account. Are more accountable to the people to be more substantive about the things that really matter to us in our daily lives about the policies that are made in our state capitol and in our nations capitol.
JAMES [interview]: Mhmm. It seems like common sense to me. [Laughs] So why is something that is so simple to implement and potentially extremely effective—why is not used more broadly, or why has it not been tested more broadly?
KYLE: Well I would say that, I think you’re absolutely right, that for you and I think it’s pretty common sense, for most people I think it’s pretty common sense. But I think there are political interests—special interests, political operatives, lobbyists, etc.—who operate at the extremes of either political party and for whom—they have benefited from electing candidates with 35 or 37% of the vote. And perhaps some of those candidates they have supported or that have helped to deliver on some of the legislation or other priorities that they wanted as a special interest group would have a difficult time getting a majority vote. Because perhaps they really cater to interests that are opposed by majority voters. So under of the current system, first past the post, they have a chance, a path to victory. Under ranked choice voting their path to victory is narrow if not impossible if they only represent narrow political interests. So there are those who represent those narrow interests who oppose this. And they have a lot of money to spend on politics and elections to try to oppose change.
But I think, you know, what we did in Maine was we just went out and talked to voters. We put our heads down, had our conversations, ignored the noise, and built a grassroots coalition of Democrats, Republicans, Independents, Greens, and Libertarians who believe that the system is broken, and that’s most of us. And we wanted to work together to improve this. And before when I talked about the nine governors races where somebody won without a majority? Our non-majority winning governors in the last 40 years, two were Democrats, two were Republicans, two were Independents. And so this is not a partisan thing. And proving government is not a liberal or a conservative idea, it’s something we can all get behind, and that was our message and that helped us communicate that to voters.
JAMES: As election results rolled in on November 8th, and the country watched the high drama of the electoral college, the ballot measures in Maine were quietly called: Question Five, An Act to Establish Ranked-Choice Voting, passed with 52% of the vote.
KYLE: Any time we can have a conversation about our expectations and what we want out of our democracy, not just for ourselves but for future generations, that’s a really constructive conversation to have. And if we really want to change the outputs, if we want a government that, you know, is less partisan, more oriented towards solving problems, that doesn’t get mired in gridlock, then we ought to think about changing the way we elect our leaders in the first place. To have a system that encourages greater collaboration.
And, you know, we are called upon to make this union more perfect. Our constitution does not say that this is a perfect union. It says that this was an effort to make a more perfect union in bringing us together. And I think that is what we are challenged as Americans to do. It’s our patriotic duty to act to make this a more perfect union.
[theme music and credits]
JAMES: Many thanks today to Kyle Bailey, Nisarg Shah, and the folks at RoboVote.org. The show was produced and edited by me, James Simenc. Our theme song is by Wintergatan.
Now we’re just a brand new baby of a podcast so please, help us grow. Subscribe to the show, leave us a review on iTunes, follow us on Twitter @DigitsPodcast, visit our website DigitsPodcast.com, and hey—tell your friends! Thank you guys for listening, and we’ll be back in two weeks with a new episode.
[theme music]
JAMES: Hey, welcome everybody to Digits, a show about numbers and us. Now I’ve always fascinated by numbers because they’re everywhere. They’re this tool that we depend on for pretty much everything we do. I mean, they allow us to quantify and make sense our world, and build machines and economies and civilizations. At the end of the day that’s what this show is: it’s me, with a microphone, following that curiosity and investigating stories about numbers and stories that numbers tell about us.
So, when I first started working on the show I didn’t expect to lead with an episode about politics. But then, long story short, um…America 2017. So here we are. We’re in the middle of this crazy, topsy-turvy political moment.
Now, in the wake of last year’s presidential election I was especially interested in conversations about process. About the systems we use to elect our political leaders. And I came across this story from last November that I hadn’t heard anything about at the time. It’s a story about the state of Maine, where a proposal was on the ballot seeking to institute an entirely new voting system in an effort to solve a decades old problem.
KYLE BAILEY: In Maine in nine of the last 11 elections for governor the candidate who won received less than a majority vote, so less than 50% of the vote, and in five of those races, so almost half of our governor elections, the candidate who won received less than 40% of the vote. And so over 60% of voters were voting for someone else.
JAMES: In 1974 James Longley won with 40% of the vote, 1986 John McKernan Jr. 40% of the vote, 1994 Angus King, 35%, 2006 John Baldacci 38%, 2010 Paul LePage 38%.
When it’s possible to win with less than 50%, less than even 40% of the vote, you can have a winning candidate that is actively opposed by 60% or more of the electorate. That is a problem.
Now this result was possible because of Maine’s voting system—called “First Past the Post.” That’s the simple, traditional method we’re all familiar with, where everybody picks their favorite candidate, you count the ballots, and whoever received the most votes wins. Which is fine with only two candidates. But once you add a third and a fourth and a fifth, all of a sudden you can win with a smaller and smaller plurality. In the extreme, it’s possible to win with only two votes if everybody else only gets one.
NISARG SHAH: So my name is Nisarg Shah, and I am a postdoctoral fellow at Harvard University’s center for research on computation and society.
JAMES: It turns out there is an entire field of study about voting and making collective decisions. It’s called social choice theory. That’s where Nisarg comes in. He’s part of the team behind the new website RoboVote.org.
NISARG: So the main idea behind RoboVote is that, look, political elections, everybody knows that they are flawed. But it’s very difficult to get a political reform. It’s very difficult to change the way an election happens at the national level. But the nice thing about RoboVote is that since we have the flexibility of deploying the best algorithms on the website, we can draw on decades of research in the AI community and we can deploy the best algorithms that people can use in their daily life. So at least at that level we can make an impact that we cannot hope to make in the political election scenario in the foreseeable future.
JAMES: RoboVote is designed to help small groups make optimum decisions, things like which company to invest in, which restaurant or movie to go to as a group, which destination to choose for a family vacation.
We make these kinds of decisions all the time. For example just the other day actually, I was out with a group of friends, like 10 of us, and we were trying to decide where to go to dinner. We stood around for half an hour on the sidewalk, totally indecisive, and finally somebody got fed up and said you know what, whatever, I’m going to the Thai place across the street. Now we were all hungry enough at that point not to care much, so we all went to the Thai place across the street. Was that the best choice for the group? Who knows. I’m not sure that “strongest voice wins” is a particularly great system. But the point is people are generally pretty bad at making that kind of decision in an optimum way.
RoboVote on the other hand is great at it. Here’s how it works:
NISARG: So what RoboVote would do is it would ask each participant to rank the different cuisines. So for example you might say that I want to go for pizza, maybe then Indian food, and maybe then Thai food. And I might have a different preference.
JAMES: The idea of ranking is at the core of RoboVote’s process. It’s all about information. If everybody only states their first choice then you know nothing about their preferences among the remaining options. And that matters because of a fundamental truth of collective decisions:
NISARG: There is no one right outcome, there is no one thing that can make everybody happy. Arguably some people are going to be happy with the final choice and some people aren’t.
JAMES: You can’t please everyone. It’s just not possible. So you need a different objective. Here’s where the rankings come in. Instead of thinking about each person’s reaction as binary, as either happy or unhappy, RoboVote sees happiness as a continuous spectrum. Now we can have a new goal. It’s not just about making the most people happy. It’s about making everyone as happy as possible.
If every user ranks their preferences RoboVote can see that, okay, your 1st choice will make you happiest, but your 2nd choice will make you happier than your 3rd, your 3rd than your 4th, and so on. And it can use that information to make a more nuanced decision.
NISARG: So each of us would tell RoboVote’s algorithms our preferences, but it doesn’t know these exact numbers. So what it does is it tries to imagine all the possible ways in which you can assign a number to these different three types of cuisines that are consistent with your ranked preference. And it does that with each participant.
JAMES: RoboVote doesn’t ask the user to specify exact values of happiness for different options because, well, that’s kind of hard to do. But ranking is easy. And the algorithm does the heavy lifting.
You can think about it like this: say we have 10 units worth of happiness to go around. I’m going to call them happions—that’s not a real thing, but if it catches on you heard it here first—so 10 happions. If I ranked Pizza 1st, Indian 2nd, and Thai 3rd, the algorithm says, okay, how can we split up these happions without contradicting the ranking? If pizza’s a strong favorite maybe it gives seven, while Indian only gives two, and Thai one. Or maybe, if I actually like the options pretty equally, it could be something like pizza 3.5, Indian 3.3, and Thai at 3.2. That’s a very different picture, but we haven’t broken any of the rules. There are an infinitely many possibilities. RoboVote looks at all of them, the entire range--
NISARG: It looks at what’s the possible sum of happiness that this option can derive, and then finds the option that would have the greatest sum of happiness no matter what these underlying utilities are. So it gives the greatest sum of happiness to the group even in the worst case.
JAMES: The worst case. That’s where the magic happens. For every outcome, it asks: what is fewest number of Happions the group could end up with? Then, whichever option has the best worst case wins.
NISARG: So they can authoritatively tell this group that, look, after looking at all your preferences, it seems that pizza is the best cuisine for you to go for. And what’s nice is that we have tested these algorithms on real data where people reported their actual happiness value for different options, we converted them to rankings, and we saw how these algorithms performed compared to other algorithms that worked with these rankings. And the actual level of happiness that these algorithms were able to achieve was significantly greater than the other algorithms.
JAMES: So aside from these kind of small scale, practical applications, what do you hope people will take away from RoboVote?
NISARG: So…I guess one outcome that we are trying to hope for is to convey to people that, look, when you have a question that you want to solve, when you have a group decision that you want to make, there are some objectively nice methods that you could use. And these methods are derived in decades and decades of research on artificial intelligence. And let’s not be afraid of them. Let’s rather embrace them, because they can make each and every of all micro decisions better in our daily life.
[“Travel Maine” commercial plays]
COMMERCIAL: This fall, make a brilliant getaway…to Maine!
JAMES: And now we arrive back in the land of lobsters and lighthouses and troubled gubernatorial elections…the state of Maine.
[Skype ringing]
KYLE: This is Kyle Bailey.
JAMES: Hi Kyle how you doing, this is James.
KYLE: Hi James how are you?
JAMES: Doing great, doing great…
JAMES [narration]: Now Kyle was the campaign manager for Maine’s 2016 Question 5 referendum.
KYLE: I moved to Maine in 2010 and experienced two of the nine elections in the last 40 years in which a candidate for governor won without a majority and just watched it all kind of fall apart, have the last three or four weeks of the election be entirely about strategic questions, and have nothing to do with issues or values or vision, and all be about I don’t want to waste my vote, who’s the spoiler, all that stuff. And I just thought I had to take some action and be involved and jumped in with this effort
JAMES: Okay, so…let’s start here. What exactly is Question Five?
KYLE: So Question Five was a citizen initiative on the ballot in 2016 in Maine that asked voters if they wanted to enact ranked choice voting to elect Maine’s United States senators, members of congress, governor, state senators, and state representatives.
JAMES: “Ranked choice voting”, also called “instant-runoff voting.”
JAMES [interview]: Can you walk us through how ranked choice voting works?
KYLE: So if there are more than two candidates in a race, which is very common in our state, you won’t have to just simply choose one, you’ll be able to rank them from your favorite to your least favorite. So your 1st choice, your 2nd choice, and so on. With ranked choice voting you can rank as many or as few candidates as you like. And on election night the town clerk still adds up all the votes to see, in this case, who is everybody’s first choice. Not who is your strategic choice based on polling but who do you actually think would be best for the job.
And if somebody gets a majority—50% plus one—of those first choices they win, right, game over, we’ve already reached a majority winner. But if no candidate has an outright majority of first choices then we look to see which candidate is in last place, who has the smallest pile of ballots in their favor? They’re eliminated. And we have a runoff between the top vote-getters. And let’s say it’s a three-way race, right? No candidate gets majority, the candidate in 3rd place is eliminated, we’re going to have a runoff between the remaining two. But instead of having to come back to the polls four weeks later with four more weeks of negative campaigning, millions of dollars spent on more political TV ads trashing and burning one another, disenfranchising overseas voters including men and women who are serving our country in military and in uniform—you could have an instant runoff.
And you do that with a ranked ballot, so you know who everyone’s first and second choice is. So those voters who liked the candidate who came in last place? We actually look at their ballot to see who is their second choice, who did they like between the remaining candidates? We pick up their ballot and move it to one of the piles for one of the top vote getters, and at that point someone will cross that majority threshold and win. So it’s a way of doing runoffs that’s more efficient, more cost effective, without the downsides of having a whole new election bringing voters back to the polls.
JAMES: Now, Ranked Choice Voting is very simplistic compared to the algorithms used by RoboVote. There are no calculations of happiness, no complex optimization techniques. But the fundamental premise is the same: if you get a more detailed ranked opinion from the voters, you can leverage that extra information to make a better decision, one that more clearly reflects the consensus opinion of the electorate.
And that’s important. Coming to a consensus. I mean, that’s the whole point of a Democracy, right? Enacting the will of the majority. And when that isn’t happening, it causes problems.
KYLE: Nowhere has that been the case truer, longer, more intensely than in Maine. We haven’t elected a governor in our state to their first term with majority support, with a mandate to govern, since 1966. In the state of Maine. So it’s pretty extraordinary, it goes back a long way, and over the decades it has led to increasing dysfunction in our state directly attributable to the fact that we elect governors—and this is just one example—but governors who don’t have a mandate to govern and lead and bring people together, and before they’re even sworn in we get bumper stickers saying, well, 61% of people didn’t vote for you, you’re not legitimate. Or the opposing legislative party says, well, we got a higher percentage of votes to elect our party in the legislature than you did, governor of an opposing party, we’re going to block everything that you want to do. And so the efforts to delegitimize the authority and mandate of that candidate whether or not they would have had a majority with a runoff system is not on the table for us, because we’re absent that. And so it creates real governing obstacles. And ranked choice voting will help to alleviate that. And make it so you cannot say, well, you didn’t get a majority. You very clearly did. You had a runoff process to get there, but this is actually the will of the people, is to go in this direction. So legislature and governor, you guys gotta sit down and learn how to work together.
JAMES: But it’s not just about majority rule. It’s also about improving the quality of political discourse during campaigns.
KYLE: You also have problems where one or more candidates are labeled as spoilers and conversations become all about I don’t want to waste my vote, who’s the spoiler? Instead of who has the best education policy? Who do I agree with on health care? You know, what’s this candidate’s vision for our state? We see that more and more in Maine in races up and down the ballot, but certainly prominent in our governor’s election and it’s such a disservice I think to our voters and to our state.
JAMES [interview]: Yeah, sure and that of course was something that was reflected heavily in our national election.
KB: Yeah, I mean you had a race this year where there were obviously two major party candidates but also a green party candidate, a libertarian party candidate, and others, and there was a lot of conversation about “don’t waste your vote” quote unquote for Gary Johnson or Jill Stein. You know, you might help elect the candidate you like the least. And you actually saw that messaging come from the two big party candidates too, encouraging their supporters not to waste their vote on a third party candidate and that certainly was a part of the conversation and people were talking about having to vote for a lesser of two evils. And so with ranked choice voting you don’t have to do that. You have the freedom to vote for the candidate you like the best without worrying you might help elect the candidate you like the least, knowing that if your favorite candidate can’t win and no candidate gets an outright majority your vote’s not wasted, it counts for your second choice.
JAMES: Now Maine would be the first state to implement this system, but there are a handful of cities in the US that already use ranked choice voting. You have San Francisco, Minneapolis and St. Paul, Cambridge, and Portland in Maine to name a few. Have we seen these kinds of positive effects on the politics of those cities?
KYLE: Well you’ve definitely seen—I’ve heard lots of stories from mayors, let’s say, the mayor of Minneapolis, Betsy Hodges, who in a pretty progressive city, she’s a progressive democrat, she ran a grassroots campaign for mayor of Minneapolis the first time. She knocked on more doors. She asked voters, “If I can’t be your first choice can I be your second choice?” She reached out, listened to more voters, talked to more voters, heard more points of view, and by the way her chief opponent said “don’t rank any candidates, just vote for me,” which was the biggest scandal that entire election because now voters had the power to express their opinions about more than one candidate and this candidate was telling them not to do so. And she ran a positive campaign, she was outspent three to one, but she worked hard, she talked to more voters, she built a coalition, and she won.
And she talks about how she governs as a progressive mayor in a pretty progressive city. Just a recent example there was an agenda that she was advocating for around expanding a set of issues that were important to the labor community, the labor movement, and there was a point where the business community said you know what, we weren’t even brought in on this, we don’t feel engaged. And she said okay, we’ve gotta stop, we’ve gotta go back and bring them to the table so they’re part of the conversation. It’s not that everybody will get what they want, but we do need all the stakeholders to participate in that. And she says that that type of governing is a direct result in part of her personality and how she operates, but also because of her experience on the campaign trail, and having to build majority coalitions and bring people with different viewpoints together. Which is exactly what you want in a governing process.
When you can win as we have seen in Maine with 35 or 37 or 38% of the vote, you don’t really have to talk to a lot of people who disagree with you on substantive issues. You can win by energizing your base, turning them out to vote, and beating the heck out of your opponents to push them just below your 35%. And when you govern you can just dip back into that well in terms of who are you accountable to, because if that’s who you needed to get elected, and who you needed to get reelected, then you’re gonna dance with them that brung you. And that’s a real problem when candidates can win with 37% of the vote or 35% of the vote. We need candidates who can win a majority and bring people together who have different viewpoints to build a winning coalition in a campaign and a governing coalition once they get to the state capitol or Washington DC.
JAMES: So the goal there is improved civility in politics.
KYLE: That’s definitely part of it. I mean, we know that voters and candidates in places that have ranked choice voting report significantly less negative campaigning, so not just voters but candidates saying I got less mud slung at me. So why is that? Let’s—you know, Portland Maine is a great example. The first time we had ranked choice voting we had 15 candidates. The second time we had three. Both the winners and the losers in those elections are supporters of ranked choice voting. Those two chief candidates in both those races had also served in the Maine legislature for a number of years. So they’d won and lost under a first past the post, this-is-the-way-we’ve-elected-people-in-Maine-for-a-long-time-system, and they talked about what that experience was like running for the legislature. So if they’d see a lawn sign for their opponent they’d skip that door and go to the next door. It’s just not a strategic use of their time in a political campaign. But when they ran for mayor with ranked choice voting and they were out knocking on doors, if they saw a lawn sign for one of their opponents they wouldn’t skip that door, they’d knock on it and talk to that voter. Because if they couldn’t be their first choice they wanted to be their second choice. They had to listen to that voter, try to find common ground, talk about the issues, try to figure out where they might jive together, and that candidate in search of second choice and maybe even third choice rankings in a crowded race. And that candidate who had that conversation who maybe wasn’t that voters first choice, maybe their second choice, knows that if they go after the candidate who was that voter’s first choice? If they trash and burn them in a mailer, if they have negative radio ads attacking them? The voter who liked the other candidate the best is way less likely to rank that other candidate who was negative as their second choice. Maybe not even rank them at all. And so negative campaigning can backfire when voters have the power to express their opinions about more than one candidate, and even punish candidates who engage in really negative campaigning.
Under the current system where you can say—and we did see this in the presidential election—vote for me because you hate them and this person over here can’t win? That goes away with RCV. You actually have to as a candidate—you can’t just say vote for me because you hate them and they can’t win—well who says they can’t win? They’re on the debate stage, they can participate without being told to sit down and shut up, they’re a spoiler, raise the issues, try to attract first choice and second choice rankings—it changes the conversation and the tone of campaigns to be much more positive and issue oriented so candidates are held to more account. Are more accountable to the people to be more substantive about the things that really matter to us in our daily lives about the policies that are made in our state capitol and in our nations capitol.
JAMES [interview]: Mhmm. It seems like common sense to me. [Laughs] So why is something that is so simple to implement and potentially extremely effective—why is not used more broadly, or why has it not been tested more broadly?
KYLE: Well I would say that, I think you’re absolutely right, that for you and I think it’s pretty common sense, for most people I think it’s pretty common sense. But I think there are political interests—special interests, political operatives, lobbyists, etc.—who operate at the extremes of either political party and for whom—they have benefited from electing candidates with 35 or 37% of the vote. And perhaps some of those candidates they have supported or that have helped to deliver on some of the legislation or other priorities that they wanted as a special interest group would have a difficult time getting a majority vote. Because perhaps they really cater to interests that are opposed by majority voters. So under of the current system, first past the post, they have a chance, a path to victory. Under ranked choice voting their path to victory is narrow if not impossible if they only represent narrow political interests. So there are those who represent those narrow interests who oppose this. And they have a lot of money to spend on politics and elections to try to oppose change.
But I think, you know, what we did in Maine was we just went out and talked to voters. We put our heads down, had our conversations, ignored the noise, and built a grassroots coalition of Democrats, Republicans, Independents, Greens, and Libertarians who believe that the system is broken, and that’s most of us. And we wanted to work together to improve this. And before when I talked about the nine governors races where somebody won without a majority? Our non-majority winning governors in the last 40 years, two were Democrats, two were Republicans, two were Independents. And so this is not a partisan thing. And proving government is not a liberal or a conservative idea, it’s something we can all get behind, and that was our message and that helped us communicate that to voters.
JAMES: As election results rolled in on November 8th, and the country watched the high drama of the electoral college, the ballot measures in Maine were quietly called: Question Five, An Act to Establish Ranked-Choice Voting, passed with 52% of the vote.
KYLE: Any time we can have a conversation about our expectations and what we want out of our democracy, not just for ourselves but for future generations, that’s a really constructive conversation to have. And if we really want to change the outputs, if we want a government that, you know, is less partisan, more oriented towards solving problems, that doesn’t get mired in gridlock, then we ought to think about changing the way we elect our leaders in the first place. To have a system that encourages greater collaboration.
And, you know, we are called upon to make this union more perfect. Our constitution does not say that this is a perfect union. It says that this was an effort to make a more perfect union in bringing us together. And I think that is what we are challenged as Americans to do. It’s our patriotic duty to act to make this a more perfect union.
[theme music and credits]
JAMES: Many thanks today to Kyle Bailey, Nisarg Shah, and the folks at RoboVote.org. The show was produced and edited by me, James Simenc. Our theme song is by Wintergatan.
Now we’re just a brand new baby of a podcast so please, help us grow. Subscribe to the show, leave us a review on iTunes, follow us on Twitter @DigitsPodcast, visit our website DigitsPodcast.com, and hey—tell your friends! Thank you guys for listening, and we’ll be back in two weeks with a new episode.